Introduction to how to use ctypes to improve Python's execution speed

This article introduces how to use ctypes to improve the execution speed of Python, and has certain reference value for everyone to learn to use python. Friends in need come and take a look.

">

Preface

ctypes is Python's external function library. It provides C-compatible data types and allows calling dynamic link libraries/shared libraries. Function. It can wrap these libraries for use in Python. This interface introduced into the C language can help us do many things, such as some small problems that need to call C code to improve performance. Through it, you can access the Windows system. kernel32.dll and msvcrt.dll dynamic link libraries, as well as the libc.so.6 library on Linux systems. Of course, you can also use your own compiled shared library

Let’s look at a simple example first. Use Python to find prime numbers within 1000000, repeat this process 10 times, and calculate the running time

import math
from timeit import timeit
def check_prime(x):
values ​​= xrange(2 , int(math.sqrt(x)) + 1)
for i in values:
if x % i == 0:
return False
return True
def get_prime(n) :
return [x for x in xrange(2, n) if check_prime(x)]
print timeit(stmt='get_prime(1000000)', setup='from __main__ import get_prime',
number =10)

Output

42.8259568214

Write a check_prime function in C language below, and then import it as a shared library (dynamic link library)

#include
#include
int check_prime(int a)
{
int c;
for ( c = 2 ; c <= sqrt(a) ; c++ ) {
if ( a%c == 0 )
return 0;
}
return 1;
}

Use The following command generates a .so (shared object) file

gcc -shared -o prime.so -fPIC prime.c
import ctypes
import math
from timeit import timeit
check_prime_in_c = ctypes.CDLL('./prime.so').check_prime
def check_prime_in_py(x):
values ​​= xrange(2, int(math.sqrt(x)) + 1)
for i in values:
if x % i == 0:
return False
return True
def get_prime_in_c(n):
return [x for x in xrange(2, n) if check_prime_in_c( x)]
def get_prime_in_py(n):
return [x for x in xrange(2, n) if check_prime_in_py(x)]
py_time = timeit(stmt='get_prime_in_py(1000000)', setup ='from __main__ import get_prime_in_py',
number=10)
c_time = timeit(stmt='get_prime_in_c(1000000)', setup='from __main__ import get_prime_in_c',
number=10)
print "Python version: {} seconds".format(py_time)
print "C version: {} seconds".format(c_time)

Output

Python version: 43.4539749622 seconds
C version: 8.56250786781 seconds

We can see the obvious performance gap. Here are more ways to determine whether a number is prime

Let’s look at a more complicated example of quick sort.

mylib.c

#include
typedef struct _Range {
 int start, end;
} Range;
Range new_Range(int s, int e) {
 Range r;
 r.start = s;
 r.end = e;
 return r;
}
void swap(int *x, int *y) {
 int t = *x;
 *x = *y;
 *y = t;
}
void quick_sort(int arr[], const int len) {
 if (len <= 0)
   return;
 Range r[len];
 int p = 0;
 r[p++] = new_Range(0, len - 1);
 while (p) {
   Range range = r[--p];
   if (range.start >= range.end)
     continue;
   int mid = arr[range.end];
   int left = range.start, right = range.end - 1;
   while (left < right) {
     while (arr[left] < mid && left < right)
       left++;
     while (arr[right] >= mid && left < right)
       right--;
     swap(&arr[left], &arr[right]);
   }
   if (arr[left] >= arr[range.end])
     swap(&arr[left], &arr[range.end]);
   else
     left++;
   r[p++] = new_Range(range.start, left - 1);
   r[p++] = new_Range(left + 1, range.end);
 }
}
gcc -shared -o mylib.so -fPIC mylib.c

使用ctypes有一个麻烦点的地方是原生的C代码使用的类型可能跟Python不能明确的对应上来。比如这里什么是Python中的数组？列表？还是 array 模块中的一个数组。所以我们需要进行转换

test.py

import ctypes
import time
import random
quick_sort = ctypes.CDLL('./mylib.so').quick_sort
nums = []
for _ in range(100):
 r = [random.randrange(1, 100000000) for x in xrange(100000)]
 arr = (ctypes.c_int * len(r))(*r)
 nums.append((arr, len(r)))
init = time.clock()
for i in range(100):
 quick_sort(nums[i][0], nums[i][1])
print "%s" % (time.clock() - init)

输出

1.874907

与Python list 的 sort 方法进行对比

import ctypes
import time
import random
quick_sort = ctypes.CDLL('./mylib.so').quick_sort
nums = []
for _ in range(100):
 nums.append([random.randrange(1, 100000000) for x in xrange(100000)])
init = time.clock()
for i in range(100):
 nums[i].sort()
print "%s" % (time.clock() - init)

输出

2.501257

至于结构体，需要定义一个类，包含相应的字段和类型

class Point(ctypes.Structure):
 _fields_ = [('x', ctypes.c_double),
       ('y', ctypes.c_double)]

除了导入我们自己写的C语言扩展文件，我们还可以直接导入系统提供的库文件，比如linux下c标准库的实现 glibc

import time
import random
from ctypes import cdll
libc = cdll.LoadLibrary('libc.so.6') # Linux系统
# libc = cdll.msvcrt # Windows系统
init = time.clock()
randoms = [random.randrange(1, 100) for x in xrange(1000000)]
print "Python version: %s seconds" % (time.clock() - init)
init = time.clock()
randoms = [(libc.rand() % 100) for x in xrange(1000000)]
print "C version : %s seconds" % (time.clock() - init)

输出

Python version: 0.850172 seconds
C version : 0.27645 seconds

总结

以上就是这篇文章的全部内容，希望对大家学习或使用Python能有一定的帮助，如果有疑问大家可以留言交流。

The above is the detailed content of Introduction to how to use ctypes to improve Python's execution speed. For more information, please follow other related articles on the PHP Chinese website!